A transformer-based model for detecting land surface phenology from the irregular harmonized Landsat and Sentinel-2 time series across the United States

Remote Sensing of Environment, Год журнала: 2025, Номер 320, С. 114656 - 114656

Опубликована: Фев. 18, 2025

Язык: Английский

---

Quantifying how topography impacts vegetation indices at various spatial and temporal scales

Remote Sensing of Environment, Год журнала: 2024, Номер 312, С. 114311 - 114311

Опубликована: Авг. 3, 2024

Satellite-derived vegetation indices (VIs) have been extensively used in monitoring dynamics at local, regional, and global scales. While numerous studies explored various factors influencing VIs, a remarkable knowledge gap persists concerning their applicability mountain areas with complex topographic variations. Here we bridge this by conducting comprehensive evaluation of the effects on ten widely VIs. We three strategies, including: (i) an analytic radiative transfer model; (ii) 3D ray-tracing (iii) Moderate Resolution Imaging Spectroradiometer (MODIS) products. The two models provided theoretical results under specific terrain conditions, aiding first exploration interactions both shadow spatial scale MODIS-based quantified discrepancies VIs between MODIS-Terra MODIS-Aqua over flat rugged terrains, providing new insights into real satellite data across different temporal scales (i.e., from daily to multiple years). Our were consistent these revealing key findings. normalized difference index (NDVI) generally outperformed other yet all did not perform well (e.g., mean relative error (MRE) 14.7% for NDVI non-shadow 26.1% areas). impacts exist spatiotemporal For example, MREs reached 28.5% 11.1% 30 m 3 km resolutions, respectively. quarterly annual deviations also increased slope. found topography-induced interannual variations simulated MODIS data. trend Tibetan Plateau 2003 2020 as slope steepened enhanced (EVI) doubled). Overall, sun-target-sensor geometry changes induced topography, causing shadows mountains along obstructions sensor observations, compromised reliability terrains. study underscores considerable particularly effects, scales, highlighting imperative cautious application VIs-based calculation mountains.

Язык: Английский

---

Forecasting corn NDVI through AI-based approaches using sentinel 2 image time series

ISPRS Journal of Photogrammetry and Remote Sensing, Год журнала: 2024, Номер 211, С. 244 - 261

Опубликована: Апрель 17, 2024

Precision Agriculture (PA) has revolutionized crop management by leveraging information technology, satellite positioning data, and remote sensing. One crucial component in PA applications is the Normalized Difference Vegetation Index (NDVI), which offers valuable insights into vigor health. However, discontinuity of optical acquisitions related to cloud cover huge load required processing time pose challenges real-time applications. NDVI prediction emerges as an innovative solution address these limitations. It allows for proactive decision-making providing accurate estimates, enabling farmers land managers plan essential agronomic activities such irrigation, fertilization, pest control, based on anticipated future conditions. This study introduces Artificial Neural Network (ANN) model incorporating NDVI, Water (NDWI), temperatures, precipitation predictive variables. The employs a novel series slicing algorithm, Boosting Adaptive Time Series Slicer (BATS), enhance input training dataset's variability, presenting with broader range examples. A 2-Bidirectional Long Short-Term Memory (LSTM) forecasting was developed predict values over short medium-term horizons. area used train, test validate ANN corresponds diverse landscape cultivated corn fields located Piemonte (NW-Italy). Results showed that estimates were accurate; considering three horizons predictions (5, 10, 15 days) RMSE resulted be 0.028, 0.038 0.050, respectively. Additionally, ablation tests proved most important variable enhancing model's accuracy NDWI, useful timesteps are four recent ones. To preliminary investigate capability operate wider different it applied entire Europe, using LUCAS dataset reference map locate fields. show 0.062, 0.083 0.105 5, 10 days horizons, methodology proposed this paper can possible alternative more ordinary approaches nowadays appears fundamental step precision agriculture where significantly improved. Future developments should explore use sequence-to-sequence ANNs development multiple spectral indices types simultaneously.

Язык: Английский

---

Monitoring Maize Canopy Chlorophyll Content throughout the Growth Stages Based on UAV MS and RGB Feature Fusion

Agriculture, Год журнала: 2024, Номер 14(8), С. 1265 - 1265

Опубликована: Авг. 1, 2024

Chlorophyll content is an important physiological indicator reflecting the growth status of crops. Traditional methods for obtaining crop chlorophyll are time-consuming and labor-intensive. The rapid development UAV remote sensing platforms offers new possibilities monitoring in field To improve efficiency accuracy maize canopies, this study collected RGB, multispectral (MS), SPAD data from canopies at jointing, tasseling, grouting stages, constructing a dataset with fused features. We developed canopy models based on four machine learning algorithms: BP neural network (BP), multilayer perceptron (MLP), support vector regression (SVR), gradient boosting decision tree (GBDT). results showed that, compared to single-feature methods, MS RGB feature method achieved higher accuracy, R² values ranging 0.808 0.896, RMSE between 2.699 3.092, NRMSE 10.36% 12.26%. SVR model combined MS–RGB outperformed BP, MLP, GBDT content, achieving 2.746, 10.36%. In summary, demonstrates that by using model, can be effectively improved. This approach reduces need traditional measuring facilitates real-time management nutrition.

Язык: Английский

---

Improving Wheat Yield Prediction with Multi-Source Remote Sensing Data and Machine Learning in Arid Regions

Remote Sensing, Год журнала: 2025, Номер 17(5), С. 774 - 774

Опубликована: Фев. 23, 2025

Wheat (Triticum aestivum L.) is one of the world’s primary food crops, and timely accurate yield prediction essential for ensuring security. There has been a growing use remote sensing, climate data, their combination to estimate yields, but optimal indices time window wheat in arid regions remain unclear. This study was conducted (1) assess performance widely recognized sensing predict at different growth stages, (2) evaluate predictive accuracy machine learning models, (3) determine appropriate period regions, (4) impact parameters on model accuracy. The vegetation indices, due proven effectiveness, used this include Normalized Difference Vegetation Index (NDVI), Enhanced (EVI), Atmospheric Resistance (ARVI). Moreover, four viz. Decision Trees (DTs), Random Forest (RF), Gradient Boosting (GB), Bagging (BTs), were evaluated region. whole divided into three windows: tillering grain filling (December 15–March), stem elongation (January heading (February–March 15). developed Google Earth Engine (GEE), combining data. results showed that RF with ARVI could accurately maturity stages an R2 > 0.75 error less than 10%. stage identified as regions. While delivered best results, GB EVI slightly lower precision still outperformed other models. It concluded multisource data models promising approach

Язык: Английский

---

Chlorophyll-a Estimation in 149 Tropical Semi-Arid Reservoirs Using Remote Sensing Data and Six Machine Learning Methods

Remote Sensing, Год журнала: 2024, Номер 16(11), С. 1870 - 1870

Опубликована: Май 24, 2024

It is crucial to monitor algal blooms in freshwater reservoirs through an examination of chlorophyll-a (Chla) concentrations, as they indicate the trophic condition these waterbodies. Traditional monitoring methods, however, are expensive and time-consuming. Addressing this hindrance, we conducted a comprehensive investigation using several machine learning models for Chla modeling. To end, used situ collected water sample data remote sensing from Sentinel-2 satellite, including spectral bands indices, large-scale coverage. This approach allowed us conduct analysis characterization concentrations across 149 Ceará, semi-arid region Brazil. The implemented included k-nearest neighbors, random forest, extreme gradient boosting, least absolute shrinkage, group method handling (GMDH); particular, GMDH has not been previously explored context. forward stepwise was determine best subset input parameters. Using 70/30 split training testing datasets, best-performing model model, achieving R2 0.91, MAPE 102.34%, RMSE 20.4 μg/L, which were values consistent with ones found literature. Nevertheless, predicted concentration most sensitive red, green, near-infrared bands.

Язык: Английский

---

Machine learning assisted prediction of land surface temperature (LST) based on major air pollutants over the Annamayya District of India

International Journal of Engineering and Geosciences, Год журнала: 2024, Номер 9(2), С. 233 - 246

Опубликована: Июль 28, 2024

Remote sensing (RS), Geographic information systems (GIS), and Machine learning can be integrated to predict land surface temperatures (LST) based on the data related carbon monoxide (CO), Formaldehyde (HCHO), Nitrogen dioxide (NO2), Sulphur (SO2), absorbing aerosol index (AAI), Aerosol optical depth (AOD). In this study, LST was predicted using machine classifiers, i.e., Extra trees classifier (ET), Logistic regressors (LR), Random Forests (RF). The accuracy of LR (0.89 or 89%) is higher than ET (82%) RF classifiers. Evaluation metrics for each are presented in form accuracy, Area under curve (AUC), Recall, Precision, F1 score, Kappa, MCC (Matthew’s correlation coefficient). Based relative performance ML it concluded that performed better. RS tools were used extract across spatial temporal scales (2019 2022). order evaluate model graphically, ROC (Receiver operating characteristic) curve, Confusion matrix, Validation Classification report, Feature importance plot, t- SNE (t-distributed stochastic neighbour embedding) plot used. On validation classifier, observed returned complexity due limited availability other factors yet studied post availability. Sentinel-5-P MODIS study.

Язык: Английский

---

A Dynamic L-System-Based Architectural Maize Model for 3-D Radiative Transfer Simulation

IEEE Transactions on Geoscience and Remote Sensing, Год журнала: 2024, Номер 62, С. 1 - 20

Опубликована: Янв. 1, 2024

We integrate the time series simulation capability of maize model within Extended L-system (ELSYS) using growth equations from a 4D and leaf breakpoint model. These models simulate emergence to male anthesis, accounting for 3D architecture during vegetative season. employ two methods achieve in ELSYS: directly use models, name ELSYS coupling (ELSYS _4Dmaize ). Alternatively, replace stem radius-leaf order function with radius-height linear interpolation transform width-length ratio constant value that varies order, thereby simulating structure, Dynamic based Architectural (DLAmaize) The DLAmaize is applied canopy reflectance simulations Discrete Anisotropic Radiative Transfer (DART) radiosity-graphics combined method (RGM), along comparison 1D Scattering by Arbitrarily Inclined Leaves (SAIL) Simulated differs significantly hotspot direction (the absolute relative difference can be up 67.4%). In addition, comparisons among radiative transfer show DART RGM yield close results. SAIL yields significant differences (e.g., 41.16% overestimate nadir reflectance) owing its assumption homogeneous canopy. enhances remote sensing dynamic vegetation canopies modeling holds promise applications.

Язык: Английский

---

Chlorophyll-a Estimation in 149 Tropical Semi-Arid Reservoirs Using Remote Sensing and Machine Learning

Опубликована: Фев. 20, 2024

Eutrophication, a global concern, impacts water quality, ecosystems, and human health. It’s crucial to monitor algal blooms in freshwater reservoirs, as they indicate the trophic condition of waterbody through Chlorophyll-a (Chla) concentration. Traditional monitoring methods, however, are expen-sive time-consuming. Addressing this hindrance, we developed models using remotely sensed data from Sentinel-2 satellite for large-scale coverage, including its bands spectral indexes, estimate Chla concentration on 149 reservoirs Ceará, Brazil. Several machine learning were trained tested, k-nearest neighbours, random forests, extreme gradient boosting, least absolute shrinkage, group method handling (GMDH), sup-port vector models. A stepwise approach determined best subset input parameters. Using 70/30 split training testing datasets, best-performing model was GMDH, achieving an R2 0.91, MAPE 102.34%, RMSE 20.38 g/L, which values consistent with ones found literature. Nevertheless, predicted most sensitive red, green, near infra-red bands.

Язык: Английский

---

Spatio-Temporal Evolution of Olive Tree Water Status Using Land Surface Temperature and Vegetation Indices Derived from Landsat 5 and 8 Satellite Imagery in Southern Peru

Agriculture, Год журнала: 2024, Номер 14(5), С. 662 - 662

Опубликована: Апрель 25, 2024

Land surface temperature (LST) and its relationship with vegetation indices (VIs) have proven to be effective for monitoring water stress in large-scale crops. Therefore, the objective of this study is find an appropriate VI analyse spatio-temporal evolution olive using LST images VIs derived from Landsat 5 8 satellites semi-arid region southern Peru. For purpose, (Normalised Difference Vegetation Index (NDVI), Enhanced 2 (EVI2) Soil Adjusted (SAVI)) were calculated. The information was processed Google Earth Engine (GEE) period 1985 2024, interval every five years summer season. triangle method applied based on LST-VIs scatterplot analysis, a tool that establishes wet dry boundary conditions Temperature Dryness (TVDI). results indicated better appreciation orchard over time, average 39% drought (TVDINDVI TVDISAVI), 24% severe (TVDINDVI) 25% (TVDISAVI) total area, compared TVDIEVI2, which showed 37% 16% drought. It concluded TVDINDVI TVDISAVI provide visualisation map crop offer range options address current future problems resource management sector areas

Язык: Английский

---